AI in the Travel Industry

With the advancement of generative AI tools like ChatGPT, we may soon see a shift in how travel is planned.

The conventional process of clicking through multiple web pages may be replaced with a more conversational approach to booking travel that involves fewer steps.

Although a few ChatGPT-powered AI travel tools aimed at helping plan the perfect itinerary and get travel advice have made their way onto the market, it seems like the big travel industry leaders are yet to implement generative AI in their existing platforms.

Expedia's CEO, Peter Kern, said that his company may try to use it to give customers a more conversational way to interact with Expedia or create more personalized recommendations by combining its data with the two types of data Expedia currently tracks; customers' purchase history and the most current pricing and availability of airline tickets, hotel rooms, and rental cars.

But that's currently still just talks for the future. Let's take a look at a few ways artificial intelligence is being used to plan travel right now.

Chris Tuttle, Co-Founder at Influence Kit, decided to try ChatpGPT's trip planning power by asking ChatGPT for help planning his upcoming family trip to Europe.

He found flights into Paris and departing from London and now needed to come up with something to do for the 8 days in between, possibly adding in a stop in Switzerland. His first prompt was;

"Act as a travel guide and help me plan my upcoming trip to Europe. I will arrive in Paris on [date] and depart from London on [date]. I'm traveling with my wife and 7-year-old daughters."

The response started with "Here is a sample itinerary for your trip. Day 1-2 Paris:" and then listed about 5 bullet points of things to do in Paris before moving on to days 3-4 in Disneyland Paris, 5-6 in the Swiss Alps, and so forth. Not a bad itinerary.

Tuttle emphasized the success of using the words "act like" in front of any prompt. It lets ChatGPT know what mindset to get into and generate a more detailed output. When he entered the same prompt without the "act like" command, it gave a much more basic output with bullet points of activities to do in each city but without suggested days or in any particular order.

He then continued to refine his travel itinerary with a series of follow-up prompts including:

"Update your suggestion to include train departure times, travel times, and arrival times."

"Include the train travel time in hours."

"Update the itinerary and remove Venice and include a day or two in Holland."

"Shorten the Disneyland visit to 1 day and add another day to London."

"Update my itinerary and remove the Amsterdam portion and extend the Alps visit."

After the follow-up prompts, Tuttle said that the chatbot got a bit lost with his original itinerary, but he had already gotten enough information to decide that he wanted to cut out the other cities he was trying to fit in and just focus on Paris and London. So he summed it up in one long detailed prompt and hoped for the best;

Act as my expert travel agent and help me create an itinerary for my upcoming trip to Paris and London with my family (which includes two 7-year-old girls). I want you to suggest family-friendly and must-see attractions in Paris and London. Include time stamps for each activity, cost, and other tips I should know about. Remember, you're my expert travel agent, so give me the best tips you can. Count the day I arrive as day 1 of my trip. I’d like to go to Disneyland Paris on day 3 of my trip and plan on staying until close. Include a train ride from Paris to London in the late morning on day 4. Include prices where relevant. I arrive on [date], at 9:05 a.m. and depart from London on [date], at 3:10 p.m. Include some activities on the day I depart.

The response was pretty great! Detailed with days, times, and even the entry prices of activities.

This does, however, raise a few questions as to the accuracy of the output. As Noam Toister, CEO of the travel platform Bookaway Group, points out, “The lack of up-to-date data is the biggest limitation – it’s hard to plan a trip without the knowledge of real-time availability and up-to-date travel options.”

Since ChatGPT's database is not yet up-to-date with information past 2021, the prices, opening times, and costs of the activities that Tuttle received were likely out of date and incorrect.

Other things that would currently be out of the reach for ChatGPT because of its data limitations are finding the cheapest flight options, most affordable hotels, currency exchange rates, current safety situations in specific destinations, and Covid policies. All things that, once updated, could make ChatGPT an excellent travel planner.

Alex Wawro details his interaction with Bing’s ChatGPT-powered search engine for travel advice. He started with a go-to question for anyone planning a trip;

"What are the best cheap hotels in Amsterdam?"

The response was clear and concise and even included links to the hotels, but after further investigation, Wawro found that the three hotels that it suggested were in fact the same hotel.

Bing did, however, also offer a block of hotel listings that included a map, links, and even hotel reviews which seemed to be a lot more helpful. Wawro then asked;

"What are some fun things to do when visiting Amsterdam?"

To which Bing answered; “I am sorry, I don't know how to discuss this topic.”  So, no help there. But by slightly altering his prompt, he was able to get a better response and a list of activities to do in Amsterdam;

"[R]ecommend me some fun things to do in Amsterdam."

So, a case could be made that Bing with ChatGPT is the better generative AI tool for travel arrangements because it has more up-to-date info, suggests follow-up questions that will easily continue your prompting, and offers website links to go along with its suggestions. That makes it easy to fact-check the suggestions it is actually making and makes it super easy to click over and actually book your trip rather than gather information. But there still seem to be a few bugs to work out.

Business travel is an almost $700 billion per year industry. So suffice it to say, AI will likely play a role in the future of business travel. It has the potential to help companies save money and time while freeing up resources for more productive activities.

Mihai Dinu of the travel expense company UiPath made a couple of predictions - with the help of ChatGPT - as to what role generative AI will play in the future of business travel.

In addition to travel planning, some of the applications involved customer service, travel assistance, language translation, itinerary changes, expense management, flight delay notifications, travel safety, and travel insurance.

The prompt he used to follow up with an airline for reimbursement of a canceled flight would save business travelers time and their company money. His prompt was;

"Write me a complaint for reimbursement because my flight KL1812 was canceled on March 6th."

But using AI for expense reporting is something that has the potential to really make waves in the business travel sector since it is one of the things business travelers complain about taking up time.

TripActions, a business-travel software startup now rebranded as Navan, says they will use AI to implement a receipt-scanning tool, which will categorize items from printed receipts into an expense report. Business travelers will be able to manage their travel bookings and expenses on one platform which will increase efficiency and decrease the amount of time it takes for them to get reimbursed.

So, like in many industries that have been embellished by generative AI, the question is; will AI put travel agents out of business? Well, the idea that the demise of the travel agent is coming is nothing new.

When online travel booking tools like Expedia and Google flights came out, you would’ve thought that the travel agent would have been rendered obsolete. Yet somehow, despite these advancements, travel agents remain an important resource for those seeking personalized itineraries and insider tips - especially in the corporate sector.

And while many of the big travel companies, both leisure, and business, have made some predictions about where generative AI could take them in the future, it still remains to be seen when we will see it happen. There are still some limitations to overcome with the accuracy and data sets before they can move forward.